Session Item

To validate NTCP models for radiation-induced fibrosis in breast cancer patients treated with whole breast RT.

We selected two Lyman models for moderate/severe (CTCAE v4.0 grade≥2) late tissue fibrosis (Fib2+) published in the literature (Alexander et al. PMB 2007, Avanzo et al. EJMP 2012). Model details are reported in Table 1. Both models are based on the Biological Effective Uniform Dose (BEUD) of the PTV DVH to include the effects of fractionation and dose inhomogeneity. The α/β ratio for dose corrections was 3 Gy. Alexander’s model considers the risk of Fib2+ at 5 years post-RT. Of note, Avanzo's model includes the possibility of heterogeneous follow-up (FU) through the use of a latency function (Tucker et al. IJROBP 2008) based on data from the EORTC trial 22881-10882.
We tested models in a multicentre EU/USA observational study cohort of breast cancer patients treated with RT after conservative surgery between 2014 and 2016. We selected a subgroup of patients with RT schedules similar to those for the model’s development: prescribed dose between 40 and 50 Gy, conventional or moderate hypofractionation, and no boost dose. Patients with post-operative fibrosis and atrophy (before RT) were excluded from the analysis. We also applied the latency function to the computation of the NTCP risk for Alexander’s model since 5 years of FU data were unavailable for all the patients.
Validation performance was evaluated through the area under the receiver-operating curve (AUC) and the calibration plot's slope and offset (calibration-in-the-large).

A total of 425 patients had RT schedules overlapping with those in the published development cohorts. The Median FU time was 24 (12-60) months. We processed 392 PTV DVHs for the analysis. Thirty-three pts (7.6%) with post-surgery side effects were excluded. Interestingly, only 55% of them continued to manifest symptoms in the late phase after RT suggesting the presence of surgical alterations that are resolved with time despite additional radiation damage.
Fib2+ was scored in 19.2% of the study group, with 15.7% and 3.6% grade 2 and 3 events, respectively. The calibration plots for NTCP models are shown in Fig1: calibration slope and offset of 1.09 and 3% for Avanzo’s model and 1.05 and 1% for Alexander’s NTCP. AUCs were 0.61 (CI +-0.04) in both validations; this index cannot be compared with the development cohorts since it was not reported in publications.

Two NTCP models for moderate/severe radiation-induced fibrosis were externally validated with very good results on a group of patients with RT regimens analogous to the ones of the development cohorts. BEUD based on the dose distribution in the PTV was a robust predictor of the risk for tissue fibrosis. Data are available to test the transferability of the models outside of the considered RT schedules, including the ultrashort breast irradiation and dose boosts.

[Study funded by the EU FP7  programme for research and ERAPerMed]REQUITE was funded from the European Union's 7th FP GA 601826.
RADprecise was funded by the ERA PerMed Network, Reference Number ERAPERMED2018-244.